Studies in the Physical Chemistry of Betaines and Related Substances

Roy GovrinShani TchernerTal ObstbaumUri Sivan. Journal of the American Chemical Society 2018 Article ASAP. Abstract | Full Text HTML | PDF | PDF w/ Li...
0 downloads 0 Views 1MB Size
Download PDF
1964

JOHN

[CONTRIBUTION FROM

THE

T. EDSALL AND JEFFRIES WYMAN,

JR.

VOl. 57

DEPARTMENT OF PHYSICAL CHEMISTRYIN THE LABORATORIES OF ~ Y S I O L O G Y , HARVARD AND THE BIOLOGICAL LABORATORIES OF HARVARD UNIVERSITY]

MEDICALSCHOOL,

Studies in the Physical Chemistry of Betaines and Related Substances. I. Studies of Dielectric Constants and Apparent Molal Volume' BY JOHN T. EDSALL AND JEFPRIES

Introduction The betaines and the amino acids occupy a distinctive place among organic substances. Both these classes of compounds are distinguished by their unusually high melting points, by their high solubility in water and insolubility in truly nonpolar solvents, and by the fact that in all cases their solutions have a dielectric constant greater than that of the solvent. All these properties find a natural explanation in terms of their structure as zwitterions. In additon to these general resemblances, however, there are certain notable differences between the two classes of compounds. Instead of the -+NH3 group found in the amino acids, the betaines contain the group -+NRlR2R3, where the R's may be almost any organic radicals, but are generally methyl groups. This structure greatly increases both the volume and the basic strength of the nitrogenous radical, and results in important modifications in the properties of the molecules. Moreover, the betaines, unlike the amino acids, are soluble not only in water but in alcohol and in mixtures of alcohol with ether or benzene. Furthermore, their dissociation constants and other properties indicate that they remain as zwitterions, and do not pass over to the uncharged form, in all of these solvents. This has made possible the present study of their dielectric behavior in solvents both of very high and very low dielectric constant, which reveals certain new relationships between the dielectric effect of zwitterions and the properties of the solvent. One highly polar molecule closely related to the betaines, N-dimethylanthranilic acid, is very soluble in pure benzene, and its dipole moment in this solvent has been directly determined, as well as its dielectric increment in other solvents, and its dissociation constants. Moreover, in the case of the betaines, in distinction from the amino acids, it is possible to obtain in pure form the uncharged molecule corresponding to a given zwitterion. Thus, to a be(1) A preliminary account of part of this work was published in J B i d . Chem.. 106,Proc. XXIV (1934).

WYMAN, JR.

taine of the type (CH3)s+NRCOO-, there corresponds the uncharged ester (CH&NRCOOCH3. We have made an experimental study of one such pair, namely, ortho benz betaine and its isomer, dimethylanthranilic acid methyl ester. The striking contrasts found between them throw considerable light on the relation of the electric charges to the properties of the molecule. In one important class of these substances, namely, the benz betaines, the electric charges on the zwitterion are closely attached to the benzene ring. Owing to the rigid nature of this structure the moments of these compounds can then be calculated with greater reliability than has hitherto been possible in other cases, and considered in relation to their dielectric effects in polar solvents. Zwitterions, like ions, produce electrostriction of the solvent, an effect which in theory should decrease as the radius of the charged group or ion increases. The increase in bulk of the charged nitrogenous group in the betaines as compared with amino acids should therefore lead to a decrease of electrostriction, and our density measurements confirm this conclusion. The electrostriction due to the betaines becomes much greater in media of low dielectric constant than in water. A fairly extensive study of this phenomenon has been carried out, and the results obtained are in general accord with theoretical expectations. Preparation of Materials Betaine [N-Trimethylglycine, (CH~)3NfCHpCOO-] was obtained from a pure sample of betaine hydrochloride (Eastman Kodak Co.) by the action of silver oxide, recrystallized four times from absolute alcohol by gradual addition of anhydrous ether, and dried forty-eight hours in a vacuum desiccator over phosphorus pentoxide; m. p. 293°.a Other betaines described below were liberated from their halogen salts and purified by the same method. Pyridine Betaine [CsH6NfCHzCOO-].-The hydrochloride was prepared by heating pyridine with chloroacetic acid.3 Heating for five hours a t about 60" gave a purer product than a shorter heating a t a higher tempera(2) WillstBtter, Bcr., 36, 584 (1902). (3) E.von Gerichten, ibid., 16, 1251 (1882).

Oct., 1935

STUDIES OF

DIELECTRIC CONSTANTS AND APPARENTMOLALVOLUMES

1965

ture. The hydrochloride was dissolved in alcohol and and the free betaine liberated and purified as described crystallized by addition of ether. The free betaine was above (m. p. 236-238'). The betaine was also prepared by the action of dimethyl liberated and purified as described above. Its behavior sulfate (4 moles) on m-aminobenzoic acid. When the reon heating was exactly as described by von Gerichten. action was complete, most o€the salt in the system was preDimethylphenyl Betaine [CBHs(CH&N+CH2COO-].The hydrochloride was prepared by heating dimethyl- cipitated by addition of alcohol; the filtrate was concenaniline with chloroacetic acid.* The reaction proceeded trated in vacuo until nearly all alcohol was removed; the much more slowly than for pyridine betaine: twelve to betaine was then precipitated as the triiodide by addition eighteen hours heating on the water-bath was required, of iodine dissolved in hydriodic acid. The triiodide was and the yield even then was less good than for pyridine converted to the iodide, as described by Willstatter and betaine. A considerable amount of an impurity with a Kahn for the triiodide of p-benz betaine. Para Benz Betaine.*~gTheiodide was prepared from deep blue color was present, which went into solution with the free betaine when the latter was liberated from the para aminobenzoic acid and methyl iodide, by essentially hydrochloride. The impurity, like the betaine, was sol- the same procedure as the meta compound. The melting uble in water and alcohol and insoluble in ether; but it point of the frei'betaine was 252-255'. proved to be soluble in acetone, or in a mixture of 90% Methods and Experimental Results acetone and 10% alcohol, in which the betaine is insoluble, and the two substances were separated by this Dielectric Constant.-The dielectric constant means. The purified betaine melted at 124-126' (in o f solutions of the betaines and related compounds agreement with Willstatter and Kahn). in various solvents was determined by a resonance Dimethylanthranilic Acid Methyl Ester [ortho (CHa)zNCsHaCOOCHs].-To a solution of 340 g. of anthranilic method previously devel~ped.~Measurements acid (2.5 moles) in 500 cc. of water containing just enough were made at 25O, a t a variety of wave lengths potassium hydroxide to allow a faintly alkaline solution, between 3 and 26 meters. In no case was there 1600 g. (12.7 moles) of dimethyl sulfate is added, with any indication of variation of dielectric constant stirring, during two and a half hours; 40% potassium bywith frequency in this range. The settings were droxide is added simultaneously a t such a rate as to maintain a faintly alkaline reaction. The temperature is held least sharp, indicating relatively greatest absorpa t about 10" during the addition of the first half; it is tion, for solutions in alcohol and solvents containthen gradually raised to 65". The resulting oily ester is ing a large proportion of alcohol. In all cases, extaken up in ether, dried over calcium chloride, and discepting a few involving solvents of relatively low tilled under reduced pressure. The yield was 150 g. (33% dielectric constant, the dielectric constant was of the theoretical amount); b. p. 139"at 16 mm. Dr. W. D. Peterson, who developed this method in Pro- found to be strictly linear in the concentration of fessor E. P. Kohler's laboratory, obtained in one prepara- solute in moles per liter, and the experimental tion a yield as high as 75%, using smaller quantities of results can be expressed in terms of a first degree material than those above given. Dimethylanthranilic acid [ortho(CHa)zNCeHaCOOH] equation involving the dielectric constant of the was obtained almost quantitatively by heating its methyl solvent eo and the slope of the dielectric constant ester with excess of water for eight hours in a reflux con- concentration curve 6. In the small number of denser.5 It was purified by recrystallization from ether cases where this was not so the deviation from and melted at 68". linearity was generally slight and the limiting Ortho Benz Betaine [ (CH&N+C~H4COO-].-Methyl value of 6 for infinite dilution could be estimated iodide was added to dimethylanthranilic acid methyl ester (1.5 moles of iodide per mole of ester) and the solution left from a graph with considerable accuracy. In to stand. The iodide of the methyl ester of ortho benz Table I the data in regard to dielectric constant betaine slowly crystallized out. The yield after one week are given on this basis in terms of €0 and 6. Cases in the cold was 15-20%, in agreement with Willstatter and in which the curves were not linear are marked Kahn. A large additional yield was obtained when the reacting mixture was allowed to stand for several weeks with an asterisk and the limiting value of 6 is followed by the value, enclosed in brackets, longer. The iodide of the betaine ester, after crystallization from reckoned from the difference between the dielecalcohol, was treated with silver oxide, whereby the free tric constant of the most concentrated solution betaine is liberated. Subsequent purification as described and that of the solvent. In the right-hand above; m. p. 228-224". Meta Benz Betaine.-The iodide was prepared from column of Table I is given in each case the range meta aminobenzoic acid and three moles of methyl iodide, of concentration studied. The contents of this following the method of Griessa as modified by Gumming,' table are shown graphically in Fig. 1. (4) Reitzenstein, Ann., 846, 328 (1903); Willstatter and Kahn, B n . , 87, 401 (1904). (5) Willstatter and Kahn, ibid., 87, 401 (1904). (6) Griess, ibid., 6, 585 (1873). (7) Cumming, Proc. Rgy. SOC.(London), 788, 138 (1906-1907).

Several of these substances have also been studied by Devoto, although only in aqueous (8) Michael and Wing, A m . Chcm. J . , 7, 195 (1885). (Q) Wyman, Phys. Rev., 86, 823 (1930).

1966

JOHN

T. EDSALL AND

JEFFRIES

VOl. 57

WYMAN,JR.

TABLE __ I ~

e0

N-Dimethylanthranilic acid

Betaine

~

*Benzene 71.1% Benzene-ethanol 51.0% Benzene-ethanol 28.66% Benzene-ethanol Ethanol 83.%yo Ethanol-water 73.0% Ethanol-water 65.37oJ,Ethanol-water 49.08% Ethanol-water 29.94% Ethanol-water , Water 89.%yo Benzene-ethanol *79.90% Benzeneethanol *68.9% Benzeneethanol *50.9% Benzen-thanol *27.6% Benzene-ethanol *Ethanol 44.370Ethanol-water

72.74% Benzene-ethanol 59.27% Benzene-ethanol 34.8670Benzene-ethanol Ortho henz betaine

Ethanol

73.71% Ethanol-water 52.34% Ethanol-water 31.70% Ethanol-water

. Water

90.11yoBenzene-ethanol 83.270Benzene-ethanol 75.70% Benzene-ethanol 61.78% Benzene-ethanol 25.45% Benzene-ethanol Ethanol

42.78% Ethanol-water

2.275 11.49 24.50 49.37 2.273 6.868 12.05 1z.91 24.36 31.41 35.96 40.09 49.37 66.05 78.54 3.205 4.913 7.427 12.19 18.15 24.42 52.29 78.54 78.54 4.579 6.501 9.90 16.33 24.26 35.70 47.37 60.43 78.54 3.191 4.234 5.825 9.27 12.74 18.63 24.42 53.50 78.54 19.97 24.41 38.71 53.21 78.54

+ 0.570

Range of concentration

d

0.002

0.18-0.39 ,144.29 - 4.21t0.04 ,154.25 - 9.4 * 0.1 ,214.33 5.1 * 0.1(5.55* 0.05) .00038-0.43 +10.7 * 0.1 ,124.17 +12.1 * 0.5 ,049-0.12 ,026-0.16 +11.2 * 0.3 ,041-0.23 +10.3 * 0.2 4-14.9* 0.3 . 080-0.12 +16.2 * 0.2 .14 .055-0.13 4-16.5* 0.6 ,082-0.16 +16.2 * 0.2 +13.6 * 0.8 .048-O. 12 $12.0 * 0.3 ,045-0.15 7.0* 0.7 ,00089-0.0042 +12.5 * 1.0(8.9* 0.2) .0016-0.017 +22 * 2 (10.9* 0.1) .0034-0.065 .0124.11 $19 * 2 (13.5* 0.2) +17.3 * 0.5(16.0* 0.2) .074-0.17 +19.3 * 0.3 (17.1* 0.3) ,042-0.18 +19.5 * 0.2 .20-0.29 +18.2 * 0.3 ,034-0.16 $18.5 * 0.2 ,114.33 +10.7 * 0.2 .020-0.046 +13.1 * 0.2 ,039-0.077 +17.0 * 0.2 ,054-0.11 $20.3 * 0.4 .0544. 12 +22.4 * 0.2 17-0.40 ,124.24 +23.6 * 0.2 .lo-0.22 $23.2 * 0.4 +21.3 rt: 0.2 ,082-0.20 .055-0.56 $18.7 =t0.2 $10.1 * 0.5 .0015-O. 0037 +12.3 * 1.0 .0065-0.017 .012-0.048 +24.2 * 0.8 f40.5 * 0.4 .042-0.073 ,048-0.089 +47.9 * 0.4 ,057-0.30 +54.0 * 0.5 ,062-0.094 +52.3 * 0 . 5 ,062-0.089 f57.2 * 0.5 +48.4 * 0.4 .032-0.084 .00078-0.0031 4-83.4* 4.0 4-77.4 * 2.5(70.0* 0.7) ,0043-0.019 +76.5 * 0.7 . ,0064-0.023 +76.1 * 0.7 ,021-0.039 +72.4 * 0.7 .011-0.043 f

- 1.2* 0.1

+

+

its methyl ester, values of the optical polarization were determined from measurements of the refractive index for sodium light made with a Pulfrich Zeiss refractometer. The index of refraction of the pure ester a t 25' was found to be 1.55743 and the density was 1.0950. This gives a polarization of 52.69 cc. The polarization of the ester was also calculated from the index of refraction at 25' of a 0.3867 molar solution in

Oct., 1935

STUDIES OF DIELECTRIC CONSTANTS AND APPARIENT MOLAL VOLUMES

benzene and the value obtained was 52.3 cc. The index of refraction of the pure acid, which melts at 68', could not well be measured; the optical polarization was therefore calculated from measurements of the refractive index a t 25' of aqueous solutions a t three concentrations with the following results: 0.1631 molar, 45.9 cc.; 0.1295 molar, 46.4 cc. ; 0.0925 molar, 45.6 cc. We may accept 45.9 cc. as a value correct to a t least 1%. Total Polarizations and Electric Moments In the case of N-dimethylanthranilic acid and its ester it is possible on the basis of the foregoing data and that on the densities given in the following section to calculate the polarizations and electric moments in benzene. The results, extrapolated to infinite dilution, are as follows: for the acid, 879 cc. for the polarization and 6.31 X e. s. u. for the moment; for the ester, 140 cc. for the polarization and 2.05 X 10-l8 e. s. u. for the moment.

1967

are probably reliable to within 1cc. and the others to within 2 cc., except in the very volatile alcohol ether mixtures, where the error may be 4 or 5 cc. Dissociation Constants of N-Dimethylanthranilic Acid and its Methyl Ester.-The dissociation constants were determined a t 25' with the hydrogen electrode, using a bubbling electrode of the Simms type.I2 Before titration dimethylan-

80

60

cg

40

Densities and Apparent Molal 20 Volumes The density of N-dimethylanthranilic acid methyl ester, measured with a pycnometer a t 25O, was found to be 1.0950, The densities 0 of the solid 0 - , m-, and P-benz betaines were determined to an accuracy of about 0.2% by flotation in - 10 0 20 40 60 80 benzene-chloroform mixtures, in c. which they are insoluble, with the Fig. 1.-Dielectric increments in relation to dielectric constant of solvent: following results: o-benz betaine, 1, p-benz betaine; 2, m-benz betaine; 3, o-benz betaine; 4, glycine betaine; 1.290; m-benz betaine, 1.299; p- 5, N-dimethylanthranilicacid; 6, N-dimethylanthranilic acid methyl ester benz betaine, 1.261. The apparent molal volumes of the various thranilic acid methyl ester was converted into its substances studied were calculated in the usual hydrochloride (m. p. 145-146', after recrystalliway from the density of the solutions, together zation) which is readily soluble in water, and the with the density of the solvents, measured with a hydrochloride was titrated with sodium hydroxpycnometer. Each substance was measured a t ide. The following values of the logarithmic two or three concentrations in each solvent studied dissociation constants (pK')were obtained CpK; (concentration of solute 1 to 4% by weight, gen- and fiK4 refer to dimethylanthranilic acid, pKA erally from 2 to 3% by weight) and the apparent to its methyl ester; concentration of substances molal volumes listed in Table I1 are the averages titrated was 0.04 molar). of these individual determinations. There was The acidity of the solution in 9Oy0 ethanol was no evidence of any significant change with concen(12) For details of method and calculation of gK' see Edsall and tration. The apparent molal volumes in water Blaochard, THISJOURNAL, 55, 2337 (1933).

1968

JOHN

VOl. 57

T. EDSALL AND JEFFRIES WYMAN, JR.

TABLEIT APPARENT MOLALVOLUMES (Cc.)IN VARIOUS SOLVENTS AT 25" I

Dielectric constant

Solvent

Betaine

Dimethyl phenyl betaine

Apparent molal volume Ortho hleta benz benz betaine betaine

*

Para benz betaine

Dimethylanthranilic acid

78.54 97.7 Water 146.8 146.5 145 141.3 134.6 70% Water-30% alcohol 60 135 133 134 50% Water-50% alcohol 47 %yo Water-75% alcohol 34 133 130 24.3 91.9 133 133 124 119 132 Absolute ethyl alcohol 70% Alcohol-30~oether 17.2 93.0 122 124 55'% Alcohob45yo ether 13.7 121 50% Alcohol-50~obenzene 12.5 89.0 137 134 30yo Alcohol-70~obenzene 7.0 131 132 Benzene 2.275 137.2 Dimethylanthranilic acid methyl ester: apparent molal volume in alcohol 159.0 cc.; in benzene 159.6 cc. TABLE I11 Solvent

PK:

PK:

PKh

Water 90% Ethanol

1.4

8.42 8.15

5.55 2.4

relation for the volume polarization p in terms of the dielectric constant e is to be replaced by the expression f~= (e

- 1)/3

(1)

so great that PKi could not be exactly determined. It probably lies between 1 and 2. From the earlier measurements of Cumming13 (in aqueous solution at 25') we calculate PIC; as 1.36, pK; as 8.64, and @IC,& as 3.70. This last value diverges greatly from our own. We beljeve our value to be the more reliable of the two, as it is based on seven different sets of e. m. f. measurements, in which the ratio of ester to ester hydrochloride was varied from 7 :1to 1 :7. The values of PK' calculated from these measurements all lay within 0.06 of the value given in Table 111. On the other hand Cumming (who employed a catalytic method) obtained shifting values which he himself regarded as not altogether reliable.

This expression would afford a ready explanation of the additive character of the dielectric constant SO apparent in the case of polar solutions and of the close correlation between 6, the dielectric increment per mole of solute, and the polarization of the solute in substances for which there is a basis for estimating the polarization, notably the aliphatic amino acids which exist as zwitterions. Hitherto, howeverJ there has always been a considerable degree of uncertainty in such estimates of the polarizations arising principally from the possibility of free rotation of the carbon atoms in the aliphatic chains, involving a change in the distance between the oppositely charged groups of the zwitterionic molecule. In the benz betaines the groups bearing the Discussion positive and negative charges are attached to the Constant'-1n the case Of polar relatively rigid benzene ring, and the separation of solutions the interpretation of the dielectric con- the charges, and consequently the moments and Offers a problem, at the pres- polarizations, can be reckoned with considerably ent time lacking a theoretical solution from gen- less uncertainty. We may in fact calculate the era1 principles. Certainly the Clausius-Mosotti separation of the charged groups in o-J lpz- and prelation is inapplicable and indeed yields results betaine, assuming the molecules to be unProbably times to' for the Polariza- distorted, quite simply from models based on data tions. On the other hand consideration of a vari- from recent x-ray and electron diffraction ety of experimental results in the light of our In each the positjve charge of the Of structure suggests that zwitterion is assumed t o located at the center the dielectric constant itself is in a general way of the nitrogen atom, and the negative charge Polarization* An midway between the two oxygem of the carboxyl Proportiond to the attempted rationalization of the situation has been group, owing to the resonance involving these. given One Of which leads to the that The locus of the carboxyl carbon atom is taken to liquids the C1ausius-Mosotti in the Of (15) Pauling and Brockway, €'roc. Nat. Acad Sci , SO, 336 (1934), (13) Cumming, Proc Roy SOC (London), 7 8 8 , 138 (1906-1907). (14) Wyman, THIS JOURNAL, 66,536 (1934) See also Greenstein, Wyman and Cohn, ibid , 57, 637 (1935)

and J Chem Physrcs, I,867 (1934); Hengstenberg and Lenel, Z Krist., 77,424 (1931); Pauling and Sherman, Proc N a t . Acad Sci , 10,340 (19343

Oct., 1935

STUDIES OF DIELECTRIC CONSTANTS AND APPARENTMOLAL VOLUMES

be on the straight line passing through the corresponding carbon of the benzene ring and the center of the hexagon, the ring itself being regarded as plane; the same is assumed also for the nitrogen atom. The various interatomic distances (in hgstroms) are taken as follows: carbon to carbon in the ring, 1.39; carboxyl carbon to adjoining carbon in the ring, 1.54; nitrogen to adjoining carbon in the ring, 1.4; oxygen to carboxyl carbon, 1.29. The angle subtended by the two oxygens a t the carboxyl carbon atom is taken as 125’. This gives 1.29 cos (125’/2) = 0.6 A. for the distance from this carbon atom to the locus of the negative charge, and this distance, owing to symmetry, is in a direction radial to the ring. The dipole distances (i. e., distances between the positive and negative charges of the zwitterions) calculated from these models are 3.23, 5.49, and 6.36 8. for 0-, m- and 9-benz betaine, respectively. In the case of o-benz betaine the distance calculated in this way would seem to be too small in view of the size of the carboxyl group and of the methyl groups attached to the nitrogen atom. If we take the carbon to hydrogen distance as 1.07, the radius of the hydrogen atom as half the hydrogen to hydrogen distance or 0.75/2, the distance from the nitrogen to the carbons of the methyl groups as 1.4,16 and assume the tetrahedral arrangement for the four valences of the nitrogen atom, we may estimate the minimum dipole distance as in the neighborhood of 3.9 8. This estimate” is based on the assumption that there is free rotation of both the carboxyl carbon and nitrogen atoms about the valences joining them with corresponding carbon atoms of the ring. It would seem, therefore, that the ortho compound must be somewhat deformed and exist in a state of strain involving a certain amount of potential energy.18 This would be expected to give rise to an effect of “steric hindrance” to the introduction of the methyl groups into this molecule, and corresponds to the fact that it is indeed much more difficult to effect the complete methylation of the amino group in anthranilic acid than in m- and p-aminobenzoic acids.lY If we adopt for the dipole distances of the u-, (16) Mark, Z. EIekfrochena., 40,413 (1934). (17) This would involve a li0 displacement of the valence bonds joining the carboxyl carbon and nitrogen atoms to the ring (IS) See Weissberger, SBngewald and Hampson, Trans. Farodor Soc , 887 (1934), and discussion by de Laszlo, p 892, who states: “In the case of 1-8-diiodonaphthalene the I atoms are spread out a t 18’ from the vertical and in ortho diiodobenzene 4O from the direction joining the C atom to the center of the hexagon.” (19) Willstatter and Kahn, Ber , 37,412 (1904).

1969

m- and p-benz betaines the values 3.9, 5.49, and 6.36, respectively, the corresponding moments are 18.6, 26.2 and 30.4 X lo-’* e. s. u. The same moments calculated from the data on the dielectric constant in water by means of Eq. (l),taking the optical polarization in each case as 52.7 (i. e., the value experimentally determined for the uncharged isomer Ndimethylanthranilic acid methyl ester) are 21.9, 31.0 and 36.4. The values are 18, 18, and 20% too high, respectively. It would seem therefore that the moments, or polarizations, calculated from Eq. 1, though probably somewhat too high, are closely proportional to the true values as far as we can estimate them. The behavior of the benz betaines as we pass to solvents of lower dielectric constant consisting of alcohol-water mixtures, alcohol, and alcoholbenzene mixtures is apparent from Fig. 1. Unfortunately it is impossible to dissolve these substances in pure benzene itself. If we make use of Eq. (1) to calculate the polarizations from these data, we obtain the results shown graphically in Fig. 2. The polarizations of course follow the 6 values quite closely and are fairly constant down to a dielectric constant of about 20. It is impossible to extend the measurements on the para compound below this point owing to its insolubility, but the polarizations of the ortho and meta compounds fall rapidly from here on with diminishing dielectric constant of the solvent. Indeed it is extraordinary that the 6 values for both these compounds (and of glycine betaine) seem to be converging toward the value 5.1 actually observed in the case of dimethylanthranilic acid in benzene. The results obtained for ortho and meta benz betaine in the solvents of lowest dielectric constant actually studied, are given in Table IV. Here we have tabulated, in addition to the dielectric constant of the solvent and the 6 value, the molal polarizations (P)and the corresponding moments ( p ) , calculated first from the ClausiusMosotti equation a t infinite dilution and then from Equation 1. In these calculations the apparent molal volumes are taken to be the same as in alcohol, namely, 133 cc. and 124 cc. for the ortho and meta compounds, respectively (they contribute at most less than 10% to the calculated polarizations), and the optical polarizations are again taken as 52.7 cc. For comparison, the results of similar calculations for betaine are included, in which the volume is taken as 91.9 cc. and the optical polarization as 35 cc.

1970

JOHN

T. EDSALL AND

JEFFRIES

TABLE IV

WYMAN,JR.

Vol. 57

with increasing concentration at concentrations in the neighborhood of 0.001 molar. Hence the Compound e (Solvent) 6 apparent moments deduced from studies at conOrthoB B 4.579 10.7 813 3725 6.05 13.2 centrations greater than this are much lower than M e t a B B 3.191 10.1 1177 3457 7.36 12.8 the true moments determined a t very great diBetaine 3.205 7.0 814 2401 6.11 10.7 lutions. Kraus interprets these phenomena as I t should be noted that in this range the values being due to association of dipoles, with decrease calculated on any basis are changing rapidly with in apparent moment, at the higher concentrations. the dielectric constant of the solvent. ‘Neverthe- It seems improbable that our data on the betaines less, however, we interpret the dielectric constant can be explained on this basis. In N-dimethylanthranilic acid in benzene there is actually a slight increase of 6 from the value 5.1 observed a t the lowest conto the value 5.5 centration 38 X obtained at a concentration of 0.43; in the case of p-benz betaine in 19.9g;b benzene-ethanol 6 is constant in the to 31 X molar. range 78 X Only in the case of betaine (N-trimethylglycine) in certain solvents is there anything resembling the effect described by Hooper and Kraus, and for this compound we have based our calculations on the limiting value of 6 for the lowest concentrations. If we accept the reality of the great reduction in moment indicated by the results in solvents of low dielectric constant and reject an explanation based on association, it remains to account for this in terms of a change in the individual molecules. The most natural explanation is probably in terms of a distortion of the valence bonds uniting the carboxyl carbon and the nitrogen atoms to the corresponding carbons of the ring. Such a distortion, involving a decrease in the dipole distance easily c. reckoned on the basis of simple geoFig. 2.-Polarizations calculated from Eq. 1, in relation to dielectric metrical considerations, would lead to constant of solvent. The point marked by an asterisk, corresponding to pure benzene, is calculated by the Clausius-Mosotti relation: 1, pbenz a decrease of electrostatic energy inbetaine; 2, mbenz betaine; 3, o-benz betaine; 4, glycine betaine; 5, versely proportional to the dielectric N-dimethylanthranilic acid. constant of the solvent. If we regard in this doubtful region, it appears clear that the this energy as that arising from two opposite elemoments of the betaines in these solvents of very mentary charges (*e) uniformly distributed over low dielectric constant are decidedly lower than two equal spheres of radius h whose centers are in water. It might be supposed that this effect separated by a distance R, we may apply the formis due to association of dipoles. Hooper and ula given by ScatchardZ1for the electrostatic enKrausZ0have recently shown for certain highly ergy E per mole. polar molecules dissolved in benzene that the 6 value and molal polarization decrease rapidly (21) See Cohn, McMeekin, Edsall and W t a i r , i b i r l , 66, 22iO P cc.,

x

lOl* Clausius- P,cc., Clausius- p x 1 0 1 Moootti Eq. 1 Mosotti Eq. I p

(?O) Hooper and Kraus, THISJOURNAL, 86,2266 (1934).

(1934).

Oct., 1935

STUDIES OF DIELECTRIC CONSTANTS AND APPARENTMOLAL VOLUMES

In this expression E denotes the dielectric constant of the medium and N is Avogadro's number. On the basis of this formula and the geometry of the molecule we can calculate the energy change Eo - E associated with a given angular distortion w of both the valences uniting the nitrogen and carboxyl carbon atoms to the ring. In the ortho and meta compounds the distortion is assumed to take place in the plane of the ring, which makes it most effective; in the case of the para compound it makes no difference in what plane the bending occurs. The results of the calculation, expressed in calories per mole, for a medium of dielectric constant unity, are given in Fig. 3. This figure shows how much greater must be the tendency for a distortion of the valence bonds in the ortho compound than in the two others and how relatively small is the tendency in para benz betaine. The question then arises how such a decrease of electrostatic energy compares with the corresponding energy increase due to valence distortion itself, Stuartz2 has estimated that the distortion of a valence bond by 10' involves an increase of potential energy of the order of 1000 to 2000 calories per mole. Although we do not know how this varies with the angle, nevertheless it is clear that in solvents of low dielectric constant there will probably be a considerable distortion and corresponding reduction of moment of the ortho and meta molecules. In water, calculation shows that such an effect would be relatively insignificant. It may well be that the greater dipole moment and the smaller tendency for this to reduce itself in less polar solvents accounts for the very much smaller solubility of p-benz betaine in solvents of low dielectric constant. It would appear, however, a t least in the case of the ortho compound, that this deformation would demand an approach of the charged groups to a distance actually less than that consistent with the bulk of these groups, discussed above. In regard to this, it is conceivable that when the forces become very great the interfering methyl groups are actually forced aside or that the carboxyl carbon atom is prevented from rotating, or both, with the result that the locus of the negative charge is brought into close proximity to the positively charged nitrogen atom; to this effect might be added the result of an actual polarization of the molecule under the influence of the strong field arising from the zwit( 2 2 ) Stuart, P h y s . R e v , 33, 1372 (1931).

1971

terionic charges of the molecule itself in a medium of low dielectric constant.

-40"

0"

-20°

20"

40"

w.

Fig. 3.-Energy change in relation to angular distortion, w , of valence bonds for the benz betaines. Negative values of o correspond to an increase of distance between the charges.

Of course the possibility remains of an intramolecular rearrangement of the betaines into their uncharged isomers, with a great resultant decrease in moment. At very high temperatures such a transformation does occur, as WillstatterZ3 has shown. Under the conditions of our experiments, however, there has never been any sign of this. The uncharged isomers (i. e., the methyl esters of the dimethylated aminobenzoic acids) are low melting substances, readily soluble in ether and benzene. When the betaines, dissolved in the solvents of lowest dielectric constant studied, were treated with ether, they were quantitatively precipitated, and no trace of ethersoluble material could be found. Furthermore, (23) Willstltter. Be?., 35, 684 (1902); ibid., 37, 404 (1904)

Willstdtter a n d Kahn,

1972

JOHN

T.EWL

AND JEFFRIES W

the betaine so isolated showed no change of melting point from that of the substance as originally prepared. For comparison with the betaines, we have studied the properties of one of these uncharged substances, namely, dimethylanthranilic acid methyl ester. This substance, dissolved in benzene, gives a 6 value of +0.570 and has a moment e. s. u. In solvents of higher diof 2.05 X electric constant the 6 value becomes negative. The results are included in Fig. 1 and show the striking contrast in behavior between this moderately polar molecule and the zwitterionic betaines. The behavior of the ester is in marked contrast with that of N-dimethylanthranilic acid itself. This substance shows throughout a high 6 value characteristic of zwitterionic structure, although in details it differs decidedly from glycine betaine, pyridine betaine, and ortho benz betaine in regard to 6. In pure benzene its moment is 6.31 X 10-'8 e. s. u. or more than three times that of the ester. The detailed character of the variation both of 6 and of the polarization (calculated from Eq. 1) with the dielectric constant of the solvent may be seen from Figs. 1 and 2. It would appear theref ore that N-dimethylanthranilic acid remains largely in the zwitterion form even in pure benzene. The study of its dissociation constants and that of its ester (Table 111) further supports the view that this substance is a zwitterion, a t least in water and in alcohol-water mixtures. According to the relations developed by EberP4 the ratio K,of zwitterions to uncharged molecules in a solution of a simple ampholyte is given by the equation Ks = (KdKE) - 1 where K1 is the first dissociation constant of the acid and K E that of its ester, bGth being expressed as acid dissociation constants. On the basis of the data given in Table I11 this yields a value of approximately 12,000 for K , in water and about 10 in 90% alcohol. There seems no doubt whatever therefore that in these solvents dimethylanthranilic acid exists predominantly in the zwitterion form. That a large proportion of zwitterions should remain in pure benzene is certainly surprising on the basis of our general knowledge of dissociation constants as a func(24) Ebert, Z.physik. Chem , 181,385(1926). See also Edsall and Blanchard, THIS JOURNAL, 56, 2337 (1933).

m ,JR.

VOl. 57

tion of the dielectric constant of the solvent.8 It may be noted that N-dimethylanthranilic acid shows very unusual properties in other respects also.% Unlike most zwitterions it is very soluble not only in water but also in alcohol, acetone and benzene, and quite soluble even in ether and carbon tetrachloride. It melts a t 70" to a clear liquid which solidifies on cooling and the solid so obtained shows all the properties of the original substance. In contrast with this, most zwitterionic acids melt between 200 and 30O0, and decompose on melting. Likewise this compound shows markedly different properties from the other ortho, meta and para aminobenzoic acids, whether the nitrogen is unmethylated, monomethylated or dimethylated. These melt a t temperatures from 140 to 180'; they are sparingly soluble in water, appreciably more soluble in alcohol or benzene; and their dissociation constants and those of their esters indicate values of K, in water which are not far from unity.n Devote% has indeed reported a 6 value of about 40 for meta aminobenzoic acid in water, while that for dimethylanthranilic acid is only 12,% but there are no essential inconsistencies here since the charged zwitterions in the meta compound possess higher dipole moments than those in an ortho compound and hence even though they do not predominate so completely over the uncharged molecules they still have a much higher effect on the dielectric constant of the solvent. It should be noted that the 6 value and the polarization of dimethylanthranilic acid calculated from Eq. 1 are lower in all solvents studied than for ortho benz betaine (see Figs. 1 and 2). It seems not improbable that this is because of the substitution of a hydrogen atom for one of the methyl groups, which permits the positive and negative radicals t o pack together more closely and results in a decrease of dipole moment. Indeed the moment in water calculated from Eq. 1 is 18.9 X e. s. u., or about 14% less than that of ortho benz betaine. 3.23 A,, the dipole dis(25) Professor Linus Pauling has suggested t o us that the hydrogen atom of the carboxyl group in dimethylanthranilic acid may form a hydrogen bond with the nitrogen of the methylated amino group. Such a structure might account for the high moment and other peculiar properties of this substance. (26) For previous discusdoas of this substance, see Cumming, Proc. Roy. SOL.(London). I S , 138 (1906-19071,and Devoto, Cazs. chim. ifol , 64,371(1934). (27) Ebert, Z. physik. CPcm. 111,385 (1926). (28) Devoto, Gaza. chim. i l d , 68, 247 (1933). (29) Devoto and Ardissone [Gaze. chim. ital., 64,371 (1934)] give the higher value, 16.7.

STUDIES OF DIELECTRIC CONSTANTS AND APPAR~NT MOLAL VOLUMES

Oct, 1935

tance calculated from the model of the molecule in the undistorted condition is 17% less than 3.9, the dipole distance of the more bulky benz betaine. In any case the very remarkable properties of dimethylanthranilic acid clearly demand further study for they are unparalleled among the ampholytes known to us. Apparent Molal Volumes and Electrostriction of the Solvent Betaines, possessing charged groups, produce electrostriction of the solvent, similar to that which has been shown to occur in solutions of amino acids.a0 The magnitude of the effect may be estimated by a comparison with similar uncharged substances, and with the aid o€ Traube’s atomic volumes. In Table V are listed the apparent molal volumes in water calculated from Traube’s theory for the substances studied, the observed volumes, and the estimated electrostriction, which is given by the difference between the observed and calculated volumes. Certain typical amino acids are also listed for comparison. In the last column is tabulated the value AE, which represents the difference in electrostriction between a given betaine and a corresponding amino acid (ortho, meta and para benz betaines are considered to correspond approximately to a, p and y amino acids, respectively). TABLE V BETAINESIN WATER Substance

Betaine 105.9 Dimethyl phenyl betaine 151.4 Ortho benz betaine 151.4 Meta benz betaine 151.4 Para benz betaine 151.4 Pyridine betaine 108.8 Dimethylanthranilic acid 135.2

Glycine a-Amino~tlericacid @-Alanine 7-Aminovaleric acid

Electrostriction, CC./mole

AE

97.7

8.2

5.1

146.8 146.5 145.0 141.3 99.6

4.6 4.9 6.4 10.1 9.2

8.7 8.4 8.0 5.8 4.1

134.6 43.5 92.7 58.9 90.0

0.6 13.5 13.2 14.4 15.9

12.7

Apparent molal vol., cc. Calcd. Obs.

57.0 105.9 73.3 105.9

..

.. .. ..

It is apparent that when hydrogen atoms, surrounding a charged nitrogen, are replaced by the more bulky methyl or phenyl groups, the electrostriction diminishes. In betaine, with three (30) Cohn, McMeekin, Edsall and Blanchard. J . Biol Chem., 100, Proc. XXVII (1933); TEXIS JOURNAL, 16,784 (1934). Reference may be made to these papers for the details of the calculation of electrostriction values.

1973

methyl groups, the decrease is 5.1 cc. out of 13.3, an average value for an a-amino acid. Substitution of a phenyl for a methyl group, as in dimethylphenyl glycine, produces a further diminution of 3.6 cc. These results experimentally observed accord closely with what would be expected from the theory of electrostriction produced by ions in solution.al The smaller a given ion the more intense will be the electric field in its immediate neighborhood and the greater the orientation and compression effects producing electrostriction of the solvent. These data suggest an approach to the solution of another problem. It has hitherto been impossible to decide whether most of the electrostriction produced by amino acids is due to the carboxyl or to the amino group since only the effect of the two groups in combination can be observed. In dimethyl phenyl betaine, however, the electrostriction produced is 8.7 cc. less than that characteristic of a amino acids. Almost all this diminution must be due to an effect on the charged nitrogen atom although the electrostric tion produced by the carboxylic group may also be slightly diminished by the introduction of the large groups adjacent to it. It seems fair to assume, however, that at least 8.7 cc. of the total electrostriction (13.3 cc.) characteristic of a amino acids is due to the amino group and, therefore, not more than 4.6 cc. to the carboxyl group. Adopting tentatively the same ratio for the two groups when the-charges are far apart and the total electrostriction is some 20 cc. per mole, we should assign approximately 13 cc. to the amino and 7 cc. to the carboxyl group. Utilizing the calculations of Vosburgh, Connell and Butler3* for the relation between electrostriction and ionic radius, we may deduce an apparent radius of 2.25 A. for the amino and 2.9 A. for the carboxyl group. In solvents of lower dielectric constant, the apparent molal volumes of the betaines are invariably smaller than in water. The same is true for salts, as was shown long ago by Carrara and L e ~ and i ~more ~ recently by the careful and exact measurements of Vosburgh, Connell and Butler.34 In the betaines the decrease in apparent molal volume is greater the greater the separation of the charged groups. Thus the value for (31) See Webb, ibtd., 48, 2589 (1926). (32) Vosburgh, Connell and Butler, J Chem. Soc., 933 (1933) (33) Cartara and Levi, (iazz cktm. itat , [II], 30, 197 (1900) (34) Vosburgh, Connell and Butler, J . Chcm. Soc ,933 (1933)

1974

JOHN

T. EDSALL AND

JEFFIUES

ortho benz betaine is 13 cc. per mole lower in absolute alcohol than in water; for para benz betaine, it is 22 cc. lower. In solvents other than water, the electrostriction of the betaines cannot be calculated directly from Traube’s atomic volumes, since the “Covolume’’ of Traube varies with the solvent employed in a manner depending largely on the particular substance studied.

I

I

I

20 40 60 . 80 Die1,ectric constant. Fig. 4:-Electrostriction of betaines and dimethylanthranilic acid in various solvents: abscissa, dielectric constant of solvent; ordinate, electrostriction in cc. per mole; @, o-benz betaine; 0, m-benz betaine; 0 , p-benz betaine; I, betaine; 0 , dimethylanthranilic acid. The values for dimethylphenyl-betaine (not shown) are almost identical with these for obenz betaine. The branch of the o-benz betaine curve marked AE refers to alcohol-ether mixtures; that marked AB, to alcohol-benzene mixtures. 0

We can, however, estimate the electrostriction directly by comparing a betaine in any given solvent with the isomeric uncharged ester in the same solvent. The difference between the two apparent molal volumes gives the contraction of the system due to the reaction (CH&NRCOOCHs + (CHa)aN+RCOO-; i. e., by definition, the electrostric tion.

WYMAN,JR.

VOl. 57

We have taken dimethylanthranilic acid methyl ester as the uncharged isomer for comparison with all the benz betaines and with dimethylphenylglycine. Its apparent molal volume in alcohol and in benzene is shown in Table 11. It will be seen that the two volumes are almost identical, and that both are about 8 cc. higher than the value (151.4 cc.) calculated from Traube’s theory for water.ae (The ester is insoluble in water, so that this determination cannot be made directly.) For purposes of interpolation we have assumed the apparent molal volume of the ester in alcohol-water mixtures to vary linearly with the weight fraction of alcohol in the solvent, from 151.4 to 159.0 cc. In other organic solvents the value 159 cc. (identical with that in alcohol and benzene) has been taken. In the case of betaine (trimethylglycine) we have no data on the uncharged isomer, dimethylglycine methyl ester. We have therelore calculated its volume in water from Traube’s theory (see Table V) and have assumed the value in alcohol and other organic solvents to be 8 cc. greater than in water by analogy. Dimethylanthranilic acid differs from its ester by one CH2 group, whose contribution to the volume of the molecule is known to be nearly constant and equal to 16.3 cc. per mole in all solvents. To estimate the apparent molal volume of the acid in the uncharged form, therefore, we have subtracted 16.3 cc. from the volume of the ester in the same solvent. The difference between this value and the observed apparent molal volume of the acid yields the electrostriction. The electrostriction of the substances studied is plotted against the dielectric constant of the solvent in Fig. 4. It may be seen from the figure that the electrostriction produced by the benz betaines is four or five times as great in ethyl alcohol (e = 24.3) as in water (e = 78.5) while that of trimethylglycine is nearly three times as great. vl’e may compare these experimental results with the effects to be expected theoretically from the calculations of Vosburgh, Connell and Butler for ions, Para benz betaine may be selected for comparison, since in this substance the charged groups are widely separated, and their mutual influence is small. The electrostriction in water is (35) Traube [Sammlung Chem. Vortragc, 4, 255 (1899)and references there given] has showu t h a t the apparent molal volumes of uncharged ortho, meta and para isomers are identical within 1 or 2 cc. (36) This is in harmony with the Bnding of Traube that apparent molal volumes of uncharged compounds are generally 8 to 13 cc greater io organic solvents than in water.

Oct., 1935

THE“SALTING-IN”

OF

HYDROGEN PEROXIDE

10 cc. per mole, of which we may tentatively assign 6 cc. to the carboxyl and 4 cc. to the methylated amino Groups producing this amount of electrostriction in water should from the tables of the above-named authors in ethanol produce 23 and 18 cc. of electrostriction respectively, or a total of 41 cc. The observed value in ethanol (Fig. 4) is 40 cc. The agreement is far better than we should have any right to expect, considering the approximations involved in the theory itself and in its application to such substances as the betaines. The results in mixed solvents are more difficult to interpret. It should be noted, however, that the electrostriction produced by ortho benz betaine is markedly greater in alcohol-ether mixtures than in alcohol-benzene mixtures of the same dielectric constant. It seems probable that this difference is correlated with the fact that ether is much more compressible than benzene. I t will be noted from Fig. 4 that the electrostriction of dimethylanthranilic acid is almost zero in water, and in all solvents is much lower than that of any of the betaines. Qualitatively these low values appear to parallel the relatively low 6 values of this substance, and suggest that its electrical constitution must differ in some significant respect from that of the betaines.

BY

ELECTROLYTES

1975

ent molal volumes have been determined for various betaines, and for N-dimethylanthranilic acid and its methyl ester in a variety of solvents ranging from water to benzene. 2. For the relatively rigid benz betaines] approximate dipole moments have been calculated on the basis of known interatomic distances. These have been compared with the values calculated from the observed dielectric increments with the aid of theoretical considerations. 3. In solvents of low dielectric constant] the dielectric increments (and probably the moments) of all the betaines studied fall far below their values in water. This effect appears to be due to actual deformation of the molecules rather than to association. 4. N-Dimethylanthranilic acid appears to exist largely as a zwitterion] not only in water and alcohol, but also in pure benzene, in which its dipole moment is more than three times as great as that of its methyl ester. 5 . The electrostriction of the solvent produced by betaines in water is less than that produced by similar amino acids, presumably due to the greater bulk of the positively charged group in the betaines. 6. The electrostriction due to the betaines is much greater in solvents of low dielectric constant than in water] and the magnitude of the effects observed is in good accord with theoretical predictions.

Summary 1. Dielectric constant increments and appar(37) The results of the ensuing calculation would not be significantly altered if we assigned 5 cc. to each group, or 7 cc. to one group and 3 to the other

BOSTON AND CAUBRIDGE MASSACHUSETTS

RECEIVED JUNE 13, 1935

[CONTRIBUTION FROM THE CHEMICAL LABORATORIES OF THE UNIVERSITY OB CALIFORNIA]

The “Salting-in” of Hydrogen Peroxide by Electrolytes BY M. H. GORIN Non-electrolytes are in general salted out of solution by the addition of electrolytes. The magnitude of this effect varies widely with the nature of the electrolyte and non-electrolyte, and in a few cases even changes in sign. However a search of the literature revealed that sodium chloride and sodium sulfate give relatively large salting-out effects with all non-electrolytes so far investigated, excepting hydrogen peroxide; Livingstonl found the activity coefficient] Y H ~ of~ hydrogen peroxide to be less than unity in aque(1) Livinpstou, THISJOURNAL, 60, 3207 (1928).

,

ous solutions of these salts. This is a “saltingin” effect of sodium chloride and sodium sulfate on hydrogen peroxide. On account of the exceptional nature of this result it was considered worth while to check Livingston’s work with sodium chloride, and to study the behavior of hydrogen peroxide with several other electrolytes. The data are presented in Table I and plotted together with Livingston’s in Fig. 1. The experimental method used was the same as Livingston’s; the distribution ratio of hydrogen peroxide between the salt solutions and iso-